Changes from previous version:
1) In iBCS-3.2 "late ABI detection" of ELF binaries on 2.6.20+ did not work
without "binfmt_coff.ko" loaded - now it does;
2) It is now possible to define personality on individual basis via the "PERFLAGS" file.
Reasons for the change:
a) "COFF clues" or "ELF marks" are sometimes insufficient to determine correct personality;
b) "ELF" executables called from a non-zero Domain will incorrectly remain in that Domain
- UW7 module will execute in Domain 3 (instead of 14) if called from an OSR5 application.
3) New Configuration Parameter "ABI_SIGRETURN".
If set to "Y" (default in iBCS-3.3+) an extra code segment will be allocated per thread
in order to restore application's context after returning form all signal handling routines.
Reason for the change:
default "vsyscall sigreturn" fails for COFF modules in kernels 2.6.18+.
4) On RHEL5 and FC6 first "sys_brk" call from an ELF module compiled on SVR4 or OSR5 did nothing
and was usually followed by "Segmentation fault" - this has been fixed.
5) OSR5 "Rewind Directory" is now recognised - on previous versions of iBCS/ABI SCO "getcwd"
library call would occasionally return ENOSYS.
6) New Inode and Device mapping functionality was implemented.
Details are in "INODEMAP" and "DEVICEMAP" files as well as in the post below:
http://sourceforge.net/forum/message.php?msg_id=4249361
.
iBCS-3.3
Logged In: YES
user_id=1748061
Originator: NO
Hi, i downloaded the file but when i unpacked it it gave an error and exited. It seems to still work fine but am a bit concerned about it. Has anyone else had this?
thanks
Logged In: YES
user_id=80373
Originator: NO
Hi Alex,
I gave a try at ibcs3.3 on 2.6.21.1 (stable) tonight, and got the following crash executing:
root@test:/usr/ads# file ./ads/bcheck
./ads/bcheck: 80386 COFF executable
here the syslog:
May 3 00:21:00 test kernel: general protection fault: 0000 [#1]
May 3 00:21:00 test kernel: SMP
May 3 00:21:00 test kernel: Modules linked in: abi_wyse abi_uw7 abi_solaris abi_isc abi_ibcs abi_sco abi_cxenix abi_svr4 binfmt_xout binfmt_coff
abi_lcall abi_util snd_seq_dummy snd_seq_oss snd_seq_midi_event snd_seq snd_pcm_oss snd_mixer_oss nfsd exportfs lockd sunrpc capability commonca
p ipv6 pcmcia firmware_class yenta_socket rsrc_nonstatic pcmcia_core lp psmouse pl2303 usbserial i2c_i801 sg i2c_core evdev iTCO_wdt intel_agp co
m20020_pci agpgart com20020 sky2 arcnet shpchp snd_ens1371 gameport ata_generic pci_hotplug snd_rawmidi snd_seq_device snd_ac97_codec ac97_bus pa
rport_pc parport snd_hda_intel pcspkr snd_hda_codec snd_pcm snd_timer snd soundcore snd_page_alloc ehci_hcd serio_raw uhci_hcd usbcore rtc_cmos r
tc_core rtc_lib
May 3 00:21:00 test kernel: CPU: 0
May 3 00:21:00 test kernel: EIP: 0060:[<c016715a>] Not tainted VLI
May 3 00:21:00 test kernel: EFLAGS: 00010046 (2.6.21.1-duo #1)
May 3 00:21:00 test kernel: EIP is at kmem_cache_alloc+0x1a/0x70
May 3 00:21:00 test kernel: eax: 00000000 ebx: 00000246 ecx: 00000000 edx: ef356000
May 3 00:21:00 test kernel: esi: 000000d0 edi: c190ba40 ebp: f8cb0530 esp: ef357e3c
May 3 00:21:00 test kernel: ds: 007b es: 007b fs: 0000 gs: 00d8 ss: 0068
May 3 00:21:00 test kernel: Process bcheck (pid: 3326, ti=ef356000 task=ef1f5a70 task.ti=ef356000)
May 3 00:21:00 test kernel: Stack: 00000000 fffffff4 00402680 c0170d84 00000000 ffffff9c 00000000 c01694ee
May 3 00:21:00 test kernel: c014efd1 00000000 00000001 00000000 00000010 00000000 f8cb0530 c01695cc
May 3 00:21:00 test kernel: bfffe868 f8c30d5c 00402680 00000000 bfffe868 0000001a 00000000 00402680
May 3 00:21:00 test kernel: Call Trace:
May 3 00:21:00 test kernel: [<c0170d84>] getname+0x24/0x60
May 3 00:21:00 test kernel: [<c01694ee>] do_sys_open+0x1e/0xe0
May 3 00:21:00 test kernel: [<c014efd1>] buffered_rmqueue+0x91/0x130
May 3 00:21:00 test kernel: [<c01695cc>] sys_open+0x1c/0x20
May 3 00:21:00 test kernel: [<f8c30d5c>] sys_call+0x6c/0x70 [abi_lcall]
May 3 00:21:00 test kernel: [<f8cbb209>] svr4_open+0x59/0x240 [abi_svr4]
May 3 00:21:00 test kernel: [<c0158f3e>] do_no_page+0x19e/0x310
May 3 00:21:00 test kernel: [<c014b8e0>] filemap_nopage+0x0/0x3a0
May 3 00:21:00 test kernel: [<c015944a>] __handle_mm_fault+0x20a/0x2a0
May 3 00:21:00 test kernel: [<f8c3003a>] get_args+0x3a/0x60 [abi_lcall]
May 3 00:21:00 test kernel: [<f8cbb1b0>] svr4_open+0x0/0x240 [abi_svr4]
May 3 00:21:00 test kernel: [<f8c3048f>] lcall7_dispatch+0x3ef/0x490 [abi_lcall]
May 3 00:21:00 test kernel: [<c01181fe>] do_page_fault+0x3ee/0x570
May 3 00:21:00 test kernel: [<f8c30cc3>] do_lcall+0x44/0x55 [abi_lcall]
May 3 00:21:00 test kernel: [<c0370000>] schedule_timeout+0x0/0x90
May 3 00:21:00 test kernel: =======================
May 3 00:21:00 test kernel: Code: e1 03 74 02 f3 a4 8b 44 24 0c 01 45 00 eb 8f 89 f6 83 ec 0c f6 c2 10 89 74 24 04 89 d6 89 7c 24 08 89 c7 89 1c
24 75 39 9c 5b fa <64> a1 04 00 00 00 8b 14 87 8b 02 85 c0 74 2e c7 42 0c 01 00 00
May 3 00:21:00 test kernel: EIP: [<c016715a>] kmem_cache_alloc+0x1a/0x70 SS:ESP 0068:ef357e3c
while on 2.6.20.x ibcs3.3 it works fine and outputs:
BCHECK C-ISAM B-tree Checker version 1.07
Copyright (C) 1981-1989 Informix Software, Inc.
usage: bcheck -ilnyqs cisamfiles ...
-i check index file only
-l list entries in b-trees
-n answer no to all questions
-y answer yes to all questions
-q no program header
-s resize the index file node size
will provide any further info if needed.
regards,
Mauro
Logged In: YES
user_id=1641112
Originator: YES
Hi Mauro,
Kernel 2.6.21 is not supported yet.
I've got ELF and COFF versions of "bcheck".
I will download "Zenwalk Linux 4.6.1" and do some testing this week.
Regards,
Alex.
Patch for 2.6.21
Logged In: YES
user_id=1641112
Originator: YES
Hi Mauro,
Try the following:
1) Uncompress KSL21.patch.gz into iBCS-3.3 directory;
2) Type in "patch -p1 < KSL21.patch";
3) "make";
This worked for my version of "bcheck".
Regards,
Alex.
File Added: KSL21.patch.gz
Logged In: YES
user_id=80373
Originator: NO
Hi Alex,
Tryed the patch right now.
It seems to work fine, until now.
entry.S is a ugly beast.
Thank you for the impressive responsiveness.
Regards,
Mauro
Logged In: YES
user_id=1073934
Originator: NO
Hi Alex.
I tried to use IBCS3.3 under kernel 2.6.18.2 (vainilla)
But I got this:
In file included from /usr/src/abi/lcall/entry.S:1:
include/asm/thread_info.h:51:29: error: asm/asm-offsets.h: No existe el fichero o el directorio
make[3]: *** [/usr/src/abi/lcall/entry.o] Error 1
make[2]: *** [/usr/src/abi/lcall] Error 2
make[1]: *** [_module_/usr/src/abi] Error 2
make[1]: Leaving directory `/usr/src/linux-2.6.18.2'
I fixed this with:
cd /usr/src/linux-2.6.18.2
touch include/asm/asm-offsets.h
I don't why appears (and because anyone else commented about this).
Regards.
Logged In: YES
user_id=1641112
Originator: YES
Hi Anibal,
If I were to use a Vanilla Kernel, I would "cd /usr/src/vanilla-2.6.18; make && make modules_install && make install".
Then I would boot that kernel and run "make" in iBCS-3.3 directory - "asm-offsets.h" is not a part of "kernel/linux source" (but is of "kernel/linux headers"), it gets generated during the "build" process.
Regards,
Alex.
Logged In: YES
user_id=1073934
Originator: NO
Ok Alex.
The problem is because I use a Debian way to build the kernel:
i.e:
make-kpkg clean
make menuconfig
make-kpkg --revision=Custom.1 kernel_image
So I get a debian package kernel-image-2.6.18.2_Custom.1_i386.deb
But if I do that, the "asm-offsets.h" file is generated and deleted each time I rebuild the kernel. :(
I know because I build the kernel first before I build the module IBCS3.3 and I always got that error.
Regards.
Logged In: NO
Ok Alex.
The problem is because I use a Debian way to build the kernel:
i.e:
make-kpkg clean
make menuconfig
make-kpkg --revision=Custom.1 kernel_image
So I get a debian package kernel-image-2.6.18.2_Custom.1_i386.deb
But if I do that, the "asm-offsets.h" file is generated and deleted each time I rebuild the kernel. :(
I know because I build the kernel first before I build the module IBCS3.3 and I always got that error.
Regards.
Logged In: NO
Hi all
I have installed ibcs 3.3 on OpenSUse 10.2 (kernel 2.6.18.2-34), compiled it. loaded as module and use to run an old cobol (rm-cobol) application. Everithing works fine, but it is impossible to print from the application. I have tried lprng and cups and none of them receive anything from the program. Same cobol configuration works fine with linux-abi and eve with ibcs 2
Any help?
Thanks in advanced.
jmanzano@aplinet.net
Logged In: NO
Hi, Alex
First of al, many thanks for your fast answer. Unfortunately i was out of office until today.
This was solved. It is a very strange case. Nothing related to ibcs, sorry. When i move the data and application from one server to another thru scp, this fail. If i move thru fpt it works fine with cups or lprng. I have no idea what it means, but once it moved thru ftp all works fine.
Kind Regards and a lot of thanks.
jmanzano@aplinet.net
Logged In: NO
Hi, Alex.
I'm running progress databasse under openSUSE 10.2 and different kernel variants (stock 2.6.18, vanilla 2.6.20 & 2.6.21) + ibcs3.3
Initially it seems to work really fine but every few minuts my progress applications stalls for some seconds and then continue (I've noticed this stalls occurs when pdflush is working / The rest of applications aren't affected).
Any idea about if this is a related ibcs problem?
Thanks for your nice work.
Regards.
miquel@acsat.net
Logged In: NO
Hi Alex.
I've compiled succesfully (KSL21 + KSL22 patched) ibcs-3.3 under vanilla 2.6.22 and under Ubuntu-2.6.22, but I'm not able to compile it under openSUSE-2.6.22 :-(
This is the error:
make -C /lib/modules/2.6.22.5-8-bigsmp/build M=/home/olaff/Desktop/abi modules
make[1]: Entering directory `/usr/src/linux-2.6.22.5-8-obj/i386/bigsmp'
make -C ../../../linux-2.6.22.5-8 O=../linux-2.6.22.5-8-obj/i386/bigsmp modules
CC [M] /home/olaff/Desktop/abi/svr4/socksys.o
In file included from /home/olaff/Desktop/abi/svr4/socksys.c:51:
/home/olaff/Desktop/abi/svr4/../include/util/revalidate.h: In function ‘do_revalidate’:
/home/olaff/Desktop/abi/svr4/../include/util/revalidate.h:30: warning: passing argument 2 of ‘notify_change’ from incompatible pointer type
/home/olaff/Desktop/abi/svr4/../include/util/revalidate.h:30: error: too few arguments to function ‘notify_change’
make[5]: *** [/home/olaff/Desktop/abi/svr4/socksys.o] Error 1
make[4]: *** [/home/olaff/Desktop/abi/svr4] Error 2
make[3]: *** [_module_/home/olaff/Desktop/abi] Error 2
make[2]: *** [modules] Error 2
make[1]: *** [modules] Error 2
make[1]: Leaving directory `/usr/src/linux-2.6.22.5-8-obj/i386/bigsmp'
make: *** [all] Error 2
Is there any known workaround?
Logged In: NO
Finally I get ibcs_3_3 succesfully compiled with openSUSE 10.3 kernel, commenting line: "error = notify_change(dentry, &attr);" in include/util/revalidate.h. Now progress starts and, apparently, work fine :-)
It seems error: too few arguments to function ‘notify_change’ is due to apparmor patches.
Can I fix this problem in a better way?
How can affect my workaround?
Thanks.
Miquel.
Logged In: NO
From opensuse-kernel list:
On Friday 07 September 2007 10:25, Miquel A. Noguera wrote:
> Hi.
>
> When I try to compile:
>
> error = notify_change(dentry, &attr);
>
> under openSUSE kernel, it aborts with:
>
> error: too few arguments to function ‘notify_change'
Well, that's because of the AppArmor patches; AppArmor requires an additional
struct vfsmnt parameter.
> (it compiles fine with vanilla or ubuntu kernels).
>
> Can you help me to get this line succesfully compiled in openSUSE kernel
> too?
I'm afraid that just getting the line to compile may not be enough. The
appropriate fix depends on what your code is trying to do. I suggest that you
check for similar uses of notify_change to get an idea what may be
appropriate, and then maybe ask again including more information such as
pointers to the code, a proposed patch, etc.
Andreas
Logged In: YES
user_id=1641112
Originator: YES
Hi Miquel,
The "error = notify_change(dentry, NULL, &attr);" worked fine for my applications.
Regards,
Alex.
Logged In: YES
user_id=1748061
Originator: NO
Hi,
has anyone been able to run this on open suse 10.3? (64 bit edition)
regards
Logged In: NO
We are using IBCS 3.3 , running SCO Foxpro 2.6. and I works fine.
But some times, once a week, the computer hangs after load abi_ldr.
Version: Debian Etch 2.6.18-4-686 SMP, original.
We test it on Fedora Core 6 with the same problem.
In Fedora the error log says:
BUG: SOFT LOCKUP DETECTES ON CPU#1!
c04051db dump_race+0x69/0x1af
show_trace_log_lvl+0x18/0x2c
show_trace+oxf/0x11
dump_stack+0x15/0x17
softlockup_tick+0xad/0xc4
update_process_times+0x39/0x5c
smp_apic_timer_interrupt+0x5c/0x64
apic_timer_interrupt+0x1f/0x24
DWARF2 unwinder stuck at apic_timer_ionterrupt+0x1f/0x24
leftover inexact backtrace
e02d586c xout_load_binary+0x0/0x134 [binfmt_xout]
memcmp+0x12/0x22
abi_personality+0x55/0x7c [abi_lcall]
__alloc_pages+0x6d/0x2aa
xout_load_binary +0x21/0x134 [binfmt_xout]
search_binary_handler+0xa4/0x260
do_execve+0x158/0x1f5
sys_execve+0x2f/0x4f
syscall_call+0x7/0xb
Any ideas ? Can you help us ?
thank you
hugo simon
hwsimon@copetel.com.ar
Logged In: NO
Baje la version IBCS.3.3 en Fedora 8.0, ahora como lo instalo, descomprimi en la carpeta /home/prueba pero no se que hacer, tengo cobol y necesito los abi como en redhat 8.0 que lo hago con INSMOD, pero ahora en FEDORA no se como hacerlo.
Marcos
which is compatible version for rhel5.8 i.e 2.6-18-308 and let me know whether any file has to be modified to compile properly. I want run foxplus 2.1.2
Last edit: mlthati 2014-08-04